A NNALES DE LA FACULTÉ DES SCIENCES DE T OULOUSE
B ODO V OLKMANN
On modifying constructed normal numbers
Annales de la faculté des sciences de Toulouse 5
esérie, tome 1, n
o3 (1979), p. 269-285
<http://www.numdam.org/item?id=AFST_1979_5_1_3_269_0>
© Université Paul Sabatier, 1979, tous droits réservés.
L’accès aux archives de la revue « Annales de la faculté des sciences de Toulouse » (http://picard.ups-tlse.fr/~annales/) implique l’accord avec les conditions générales d’utilisa- tion (http://www.numdam.org/conditions). Toute utilisation commerciale ou impression sys- tématique est constitutive d’une infraction pénale. Toute copie ou impression de ce fi- chier doit contenir la présente mention de copyright.
Article numérisé dans le cadre du programme Numérisation de documents anciens mathématiques
http://www.numdam.org/
ON MODIFYING CONSTRUCTED NORMAL NUMBERS
Bodo Volkmann (1)
Annales Faculté des Sciences Toulouse
Vol I,1979,
P. 269 à 285(1 )
Mathematisches InstitutA,
UniversitätPfaffenwaldving 57,
D 7000Stuttgart
80 -Allemagne
Fédérale.Resume : Soient
p~
... des entierspositifs,
ecrits en baseg > 2,
tels que le nombre reel a =p2
...soit normal. On démontre que
c~* = p * ... ,
oùp~ = p~
+j. ,
EIN*, log j~
=o(log p.),
est aussinormal,
théorème
qui
résout unproblème pose
par T.SALAT.
Parmi d’autresrésultats,
on donne unegeneralisation
auxmesures arbitraires de distribution
g-adique, x,
telles que x( ~ 1 ~ )
= 0.Summary :
Ifp~
... are naturalnumbers,
written to baseg > 2,
such that the real numberQ
g
g .1
12
... IS normal, It IS proved t at Qg
g .1 2
Witi
=i i
i ’i
i ’ ogi
i = 0 ogpi),
IS a sonormal. This theorem solves a
problem proposed by
T.SALAT. Among
otherresults,
ageneralisation
toarbitrary g-adic
distribution measures x with x( ~ 1 ~ )
= 0 isgiven.
1. PROBLEM.
Let
p2,
... be an unbounded sequence of natural numbers whichdiverges
toinfinity
and letB1 ~ B~,
... be thecorresponding
blocks ofdigits
relative to a fixed(integral)
baseg > 2 ; furthermore,
we suppo-se that the real number
(1 ) )
be normal. Mr T.
SALAT
has raised thequestion (2)
whether the numbera*,
obtained in the same way but with eachpi being replaced by pi = pi
+1,
is also normal.(1 ) )
Thepre-index
g denotes ag-adic expansion.
(2)
Communicated to the authorby
Mr F. SCHWEIGER.This
question
appears to be well motivated inasmuch as the answer is affirmative in the «classical»cases of normal numbers constructed in the form
(1) :
a)
ForChampernowne’s
normalnumber (cf. [1] ] ),
defined withg =10
andpi
=i,
the modified num-*
ber is a = 10
a -1, normality being
obvious.b)
For the normal numbers constructedby
the method of DAVENPORT andERDOS [3] ,
wherep. = p(i)
for any non-constantpolynomial
p which mapsIN*
into(1 ),
the modifiednumber a*
satisfies thesame
assumptions
and is therefore normal.c) Similary, normality
ofQ *
followsimmediately
if a is constructedby
the method of COPELAND andERDOS [2]
which allows(p;)
to be any sequence with more thann 1- E
elements notexceeding
n for everyE > 0 and all n >
no(E ),
inasmuch asagain a*
satisfies the same condition.It is the main purpose of the
present
paper to show that the answer to the statedquestion, in aslight- ly
moregeneral form,
isalways
affirmative(Theorem 1).
Moregenerally,
it will be shown for which distributionmeasures x,
assuming
that a have thedigit
distribution x, the same isalways
true fora* (Theorem 4).
Further-more, we
give
some classes ofcounter-examples
which show that in Theorem 1normality
cannot bereplaced by simple normality.
Inparticular,
Theorem 3 describes allpossible pairs
ofdigit frequency
vectors which thepair
of
numbers a,
a may possess.The work contained in this paper was done
during
the author’s stay at Bordeaux in thespring
of1978,
and he wishes to express his thanks to the U.E.R. deMathématiques
etd’l nformatique
for all thehospi- tality
andcooperation
received.2. NOTATION.
2.1. - With every
integer
m E fN we associate the block B =l3(m)
of itsg-adic digits,
written in the cus-tomary order. We consider the set
Big
of allg-adic blocks, including
the emptyblock,
to be denotedby F.,
as asemi-group
with respect tojuxtaposition,
thusregarding j3
as amapping
from IN intoBg.
Let Bog
be the subset of obtainedby omitting F03C6
and all blocks oflengths >
1 whichbegin
with the
digit
0. If~ g
isreplaced by ,~g,
themapping @
becomesbijective.
2.2. - For any block F we denote the
length by
II FII , defining
IIF~
II = 0.2.3. - Blocks of the form F F ... F
(n times)
shall be denotedby F~n).
If F is thesingle digit g-l,
weshall also write
(g-1 )~n)
=(1 )
We denote the set of natural numberby (N*, letting (N* U ~ 0 ~ _
2.4. - If
F ~ F~
we letp(F)
and a(F), respectively,
be thelargest integers ~
0 such that F can be written in the form F = =F"(g-1 )(o ) (see 2.3) F,
withF’,
F"g (where evidently
either F’ = F orF" = F).
2.5. - For any block F ~
F03C6,
and any real numberx E (0,1] let AF(x,n)
denote the num-ber of
copies
of Foccurring
within the first ndigits
of the(non-terminating) g-adic expansion
of x, withAF ~ (x,n)
= n.Similary,
if let denote the number ofcopies
of F which are embedded in B.Thus, normality
of a may beexpressed by
means of theequations :
The number a is called k-normal if
(2)
is satisfied for all F with 11 FII = k,
and1-normality
isusually
calledsimple normality.
2.6. -
Any
block F E~g
determines the «basic interval» of numbers xE ( 0,1 ] whose g-adic
expan-sion
begins
with F. For the sakeof simplicity
we shall denote this intervalby F also, letting F~=(0,1 ].
2.7. - For any set L c
IN
letL(n)
denote the number of elements notexceeding
n(n = 1,2,...).
3. MAIN RESULT.
We can now state our main result :
THEOREM 1. Let
pi, Bi (i
=1,2,...)
and a be defined as in(1),
letJl,l2,"’,~i
EIN,
be numbers such thatlog j.
=o(log p.), p. = P ;
=Bi (i
=1,2,...).
Thennormality
of aimplies normality
of the numbera)
We define the numbers MB;
!! =q;,
!! !! =q~ s~
=q~
+ ... +q;, s~=
+ ... +z~
= !!j~(j;)
!!and the sets S =
~ ,s~,... ~ S* = ~ s~ ,s~... ~,
thushaving
Consequently,
sincep. i -~
we obtains. i+1 - s. i =
1 7 ~ ands* i+1 - s/-
whichimplies
that(1 )
~)
See 2.7.b)
If i issufficiently large
such thatqi
>zi
we rewriteBi
in the formwhere
II Bi3 II
=z.. Furthermore, let ~ 1 ) k. =
and rewriteB’. in
the formB. = Bi2
withII Bi2
Thus we have obtained a
representation
where
Bil
is either empty or has a lastdigit
different fromg-1,
andBi2
is eitherempty
or consist of(g-1 )’s only.
In this notation the transition from
p~ to p~ + j~ changes
the blockB. into
a block of the formwith the
following properties:
IIB*i3
II = IIBi3
II ,according
as to whether or not theinequality
holds
g
= 10, B; = 6089995301, j ~ i = 4711, k, = 3, z; = 4, 6090000012, B. i1
=608, B.2 i
=999, B. i3
=5301,
=
609, B~ i2
=000, B.~ i3
= 0012.c)
If the n-thdigit
of a occurs within the blockB;,
i.e. if nsi,
then the«corresponding»
digit
ofa*
has the index n’ for whichs;
- n’ =s; -
n. Inparticular,
we have n’ > n for all n, and theimage
setof
fN*
under thismapping
hasdensity
one, itscomplement being
contained in the sets~
+1, s2
+1,....
d)
Therepresentation .
induces an obvious
decomposition
of the setflU*
into threedisjoint
sets11, I 2
and)~ where Ik
consists of those indices whoseposition
isoccupied by
one of the blocks w(k -1,2,3).
We need thefollowing
twolemmas.
Proof. From the definitions of
qi and ji
we have :Hence,
if n issufficiently large
and nsi,
it follows thatwhere the relation
(5)
has been used.c p
Proof. If e >0 is
given,
we choose aninteger
C >0 such thatc .
I2
be the set of those n~ , , , , , g p ~
which are associated with one of the last C
digits
in some block ThenIx2
consists of chains oflengths
Cof consecutive
integers
each of which liesstrictly
between two consecutive elements of S.Hence,
for any nand
thus, by (5),
Let
I2l = 12 1 ( 2 Q,
then every m E is an index at which in theexpansion (1 )
a copy of the block(1) G2 begins.
Therefore,
it follows from(2)
thatfor all
sufficiently large
n.(~)
See 2.3.The assertion is
implied by (10) and (11 )
sincee)
In order to establishnormality of a* by
means of theequations analogous
to(2)
it suffices to let n tend to *’infinity through
the elements of *IN*’ (see step (c)).
We consider a fixed block * F ~=F~
and an indexn’ E
and
we subdivide theAF(a ,n’) copies
of F among the first n’digits
of a into three subsets :1)
LetA1 (a*,n)
be the number ofcopies occurring within,
but not at the end of some blockB.. By
virtueof
(9),
thesecopies correspond
tocopies
of F in the blocksBil.
Hence2)
The numberAr-(o! ,n’)
ofcopies
of F which occur at the end of some block8.1 (for
thesecopies
thereis no
corresponding
copy of F in theexpansion (1))
is boundedby S* (n’)
+1,
therebeing
at most one such copy in each blockB.. Hence, by (5),
3)
Each of theremaining copies
of Foverlaps
with at least one of the blocksBi2
orBi3. Hence,
their num-ber is
and thus it follows from Lemmas 1 and 2 that
By combining (12), (13)
and(14)
we obtainand
therefore, by (2),
we haveAdding
theseinequalities
for all blocks F of the samelength
q andobserving
thatwe find that
equality
must hold in(15)
for every F. Asimple
argument shows now that the sameequations
arevalid for the lower
limits,
and thus the assertion follows.4. REMARKS. ,
4.1. - The assertion we have
proved
ceases to hold ifnormality
isreplaced by simple normality.
Forexample,
letg > 2, B. _ (g-2)~~) (g-3)~~)
...0~~) (g-1 )~~) (i
=1,2,...), j j~
= 1 for all .. i. Then theintegers
~B~) i
areuniquely determined, satisfying p 1 p 2
.... Inasmuch as every blockB. contains i zeros,
iones etc.., the number a
= . B~ B~
... issimply
normal to base g. But thenumber a * = g . Bi B;
... involves theblocks
A
simple
calculation shows thatHence,
the numbera*
fails to besimply
normal.4.2. - The
example just given
may be altered in such a way that the number a does not even possessan
asymptotic distribution,
as we shall demonstrate in the case g = 2.Indeed,
letand define blocks
Bi (i =1,2,...)
asWriting again p; = 0 (B.),
the numbersp.
areuniquely
determined(as
each blockB. begins
with thedigit 1),
and
they
form a monotonic sequence since IIBl II
IIB2
II ....Hence,
the number a= 2 . Bl B2
... satisfiesthe conditions of Theorem
1,
except that instead ofbeing normal,
it issimply
normalonly.
For the modified number
a~ . B* B;...
one has)
Now, let 03A3 k! = nj (j=1,2,...) and
k=1
These parameters
satisfy
theasymptotic equations
and
It will be convenient to rewrite the number
ain
the formwhere
Each of the blocks
f 1 f 2
...f
, in(16)
ends with the sub-block 1(2j) ; hence,
thecorresponding
block in the ex-pansion
ofa*, consisting
of thefirst m2j digits,
ends with a sub-blockQ2j
in which all theB~ satisfy
the secondalternative of
(17),
thushaving A~
=2. 2~ i
Therefore,
and thus
Consequently, using (18)
and(19),
we obtainOn the other
hand,
a similarargument
shows thatwhere the first alternative of
(17)
has beenapplied.
Therefore,
Since the
opposite inequality
is a trivial consequence of(17),
we have*
which shows that no limit
frequencies
for thedigits
of a* exists.4.3. - Another
modification
of theexample given
in 4.1. demonstrates that the number a can be sochosen that it remains
simply
normal to base 2 buta*
hasgiven
lower and upperasymptotic digit frequencies, subjet only
to the conditionwhich is
prompted by
the fact that a issimply
normal and theasymptotic frequency
of zeros cannot be decreasedby
the transition from a toa*.
We restate thisproposition
as thefollowing
theorem.THEOREM 2. Given real numbers r~ and
~’ satisfying
there exists a sequence
pi
-~ ~ of natural numbers such that the number a constructed to base 2 in the sense of(l)
issimply
normalbut, letting (1 ) j. i =1
for alli,
the numbera*
satisfies theequations
Proof. We consider the same
number 7
as defined in(16)
but wechange
the definition of the blocksBi
as fol-lows :
for j =1,2,...
we letand if
nj’
we defineThen the
integers pi
=~3 (B.)
areeasily
seen tosatisfy
the conditionPi i -~ ~
and the number a= . 2 B1
1B~ ...
is
simply
normal sinceThe modified blocks are
(1)
The theorem is also true forarbitrary sequences ji
withlog ji
= 0(log pi)
but theproof
would be more com-pi icated.
A 0 (a ~m)
Using again
the notation(20)
we findby
astraight-forward
calculation that the lower limit lim is~
mobtained as m runs
through
theending positions
of the blocksQ2~ (j =1,2,...).
These values of mcorrespond
tothe
end-points
oflong
chains of blocks for which the first alternative of(21 )
isvalid,
and sincein this case, the result is
Similary, using
the second alternative of(21 )
and the blocks thushaving
one obtains
4.4. - The result
just proved
can begeneralised
to anarbitrary
base g. Instead ofdoing
this we shallreplace
the condition that the number a besimply
normalby requiring
that a anda*
havegiven digit frequencies.
THEOREM 3. Let
g >
2 be aninteger
and letg-adic digit frequency
vectors(1) 03BE
=(03BEo,03BE1,
,..., 03BE _ )
,§
_(03B6*o, 03B6*1 , ..., 03B6 g-1 )
begiven.
Then there exists a sequencep2
... suchthat, letting ( ) j i
; =1 forall i the numbers a and
a*
in the sense of Theorem 1 havedigit frequencies ~’
and~’ *
if andonly
if(3)
COROLLARY. . The
equations (23) imply
the relationProof. a.l. Assume that numbers a and
a*
whith the statedproperties
exist. In order to compare withA (B.) (i = 1,2,...)
wedistinguish (in
the notation of section3)
thefollowing
five cases(4) :
1 *
g-l g-l
(1 )
I.e. real numbers~’k =1 .
k=0 k=0
(2)
The theorem is also true forarbitrary sequences ji
withlog ji
= 0(log p;)
but theproof
would be more com-plicated.
(3)
It should be observedthat,
for g =2, (24)
is trivial and(23)
is void.(4) Observing
that in case 3 one hasa(B.)
+p(Bil)
IIBi
II , we find that this list of cases iscomplete
andnon-overlapping.
For g =2,
cases 1 and 4 do not occur.On
modifying
constructed1) Q (Bi) = 0 , p (Bi) = 0 (example : 19805)
2) a (Bi) = 0 ,
p >0(example : 19800)
3) Q(Bi) >0 ,
P(Bil )
> 0(example : 900999 ;
=900) 4) II Bi
II >a (Bi)
>0,
=0(example : 108999, Bi~
=108)
5)
IIB~
II =a(Bi) (example : 99999).
Computing
the blocksB = ~i(~i ~ (Bi)
+1 )
we obtainFurthermore,
we findA
simple
argument,using
the relation IIB* i
II =q.~ i -~ ~ (see (4))
to handle the terms -1 in case2,
now showsthat
Hence,
condition(22)
is necessary.~y.2. - - To prove the
necessity
of(23)
we may assume that g > 2. Let k be a fixeddigit
with1 k
g-2. Then, considering again
the five cases introduced ina.1),
we findalways
thatAk (Bi)
= 0or ±
1,
from which theequations (23)
can be deduced.b.
Conversely,
let us assume thatvectors 03B6 and 03B6*
with the statedproperties
begiven.
Then weconstruct a
pair
of numbers a,a by modifying
the construction of Theorem 2 as follows : we nowintroduce,
forj
=1,2,...,
the functionsnoting that,
in view of(24),
If nj-1 i nj , we define the block Bi as
letting again
a= g . B~ B~
...,B~= (B.)
+1),
andIt is easy to show that the number a has the
digit frequency vector § , using
the relations IIBi
II= j -~
oo,i (h = 0,...,g-2)
andUj
+( ~’ g-1 * + ~ ~ - ~ 0 o)j = ~ - g 1 j,
the last onebeing
obtainedby
meansof
(24).
Similary,
inasmuch asan obvious
computation, using (27)
and the relation* *
proves that the number a has the
given digit frequency vector t
.5. GENERALISATION
Theorem 1 may be
interpreted
asstating
that if the number a hasLebesgue
measure X as itsg-adic
*
distribution,
the same is true for a .Hence,
one may ask whether there are distribution measures other than Bhaving
the same invarianceproperty.
Thisquestion
is answeredby
thefollowing proposition.
THEOREM 4. Let x be a
g-adic
distribution measure. Then any number c~ constructed as in Theorem 1 andhaving
theg-adic
distribution(1 )
x , , willyield
a numbera*
with the same distribution if andonly
if(2)
x
( 1 ~ ) = 0,
i.e. if thepoint
1 is not an atom of x .(1 ) ) I.e.,
for everyblock F,
lim =x (F)
in the sense of 2.6.o n n *
( )
It should be noted that this condition is violatedby
theexamples
discussed in section 4 whenever c~ and ahave different distributions.
Proof.
a)
Let x be agiven g-adic
distribution measure, i.e. a. Borelprobability
measure on[0,1 ] which
isinvariant under the
g-adic
shift operatorT g
x= g x ~ , satisfying
theassumption x ( ~ 1 ~ )
= 0.Parts
a), b), c)
andd)
of theproof
of theorem 1 remainvalid,
except that in theproof
of Lemma 2the
following change
is now necessary. Since(~ ) the
intersection of the basic intervalsG Q
isthe
assumption x ( { 1 } )
= 0 isequivalent
toHence,
if e > 0 isgiven,
there exists an Q such that x(GQ)
e . The remainder of theproof
isunchanged,
ex- cept that theinequality (11 )
has to bereplaced by
The fact that
a*
has distribution x is then establishedby
the argument of parte)
in theproof
of Theorem1,
1
with the term - in
(15) being replaced by x (F).
b) Conversely,
let x be ag-adic
distribution measure with x({ 1})
= ~ > 0. We shall show that there exists a sequencepi
withpi
l’ oo suchthat,
in the sense of thetheorem,
a has the distribution x buta*
does not.
b.l. - As the first step we consider a number
a1
with the distribution x as constructed(2) ) by J.
VILLE[6], following
theexposition
of A.G. POSTNIKOV[5], Chapter
III. Given any sequencee
1 >e 2 >..., E k ~ 0,
this construction definesrecursively
blocksLo,
..., in such a waythat, letting
and II
Lo L~
...Lk
II =nk,
any block F =~F~,
II F IIk,
satisfies theinequalities
for all n >
nk,
Thelength
IILk
II may, at each stage of theconstruction,
be chosenarbitrarily large
in relation to the II!!. Hence,
we may choose eachnk+1 (k= 1,2,...)
solarge that,
for all F with II F IIk,
(~)
See 2.3 and 2.6.(2) )
Anelegant
construction of suchnumbers, using graph theory,
wasrecently given by
H.KUHNLE [4] .
For the same reason we may
furthermore
assume thatFurthermore,
werequire
that the firstdigit
ofLo
be different from 0.6.2. -
By (29)
we haveHence,
anapplication
of Dirichlet’s drawerprinciple
shows that for each k there exists an arithmeticprogression
{ rk’ rk +k, rk +2k, ...}
, 1k,
which contains more than1 k R(k) numbers
n such that some copyof G
ends at the n-th
digit
of the blockLk+ 1.
Let be the block obtained from
(k
=0,1,2,...) by
acyclic
shiftby k-rk places
to theright.
I n view of(30)
it ispossible
to rewrite each block(k = 1,2,...)
in the formwhere
II
1 II = ... =II Y k h
II = k and at least -R(k)
of these blocks areequal
toGk,
to be called relevantk k
copies.
Now,
let a numbera2
be defined asIt is easy to see that
a2
also has thedigit
distribution x .Indeed,
if ~denotes the set ofplaces occupied
in theg-adic expansion
ofa~ by
the first and the lastdigit
ofL~,
the first two and the last twodigits
ofL2,etc,
thenit is evident
that,
if a block F with II F II = k > 0 isgiven,
the transition fromc~i
toa2
canonly
create ordestroy copies
of F to theright
of the indexnk
ifthey begin
or terminate atplaces belonging
toHence,
wehave,
for all nnk,
and
since, by (31 ),
=o(n),
it follows thati.e.
a2
has the distribution x .b.3. - The blocks
(k = 1,2,...)
arechanged
as follows : we define blocksthus
replacing (33) by
and then
letting a3 = g . L’2 L3 L4 ..,
.The first
digits
of all blocksYki
determine a set V= ~ Vi, v2, ... ~
of indices for whichvm = k
wheneverthe index
vm
occurs within the blockLk. Hence, by (31 ), V(m)
=o(m)
and thusa3 again
has thedigit
distribu-tion x .
Furthermore,
it follows from the definition of 1 that(with (~k+1 ) - 1 2).
6.4. - -
Finally,
we renumber the blocksYki (k = 1,2,... ; i=1,...,hk)
withoutchanging order,
asB1,B2,w,
thusrewriting a3
as a= g . B~ B2
.... Then the number a satisfies theassumptions
of the theorem since :1)
it has thedistribution x , 2)
none of theblocks Bi begins
with0,
hence theintegers pi = 03B2-1 (Bi)
areuniquely determined, 3) Pi
as IIBi
IIFurthermore,
each relevant copy of a blockGk (k = 1,2,...)
forms one of the blocks
Bi.
We shall also use the fact that theassumption log ji
=o(log p~) implies
the relation~ ~ ~
~.3. - For the
corresponding
number o;= . Bi B~ ...
we also use the block decomposition
where !s obtained from
L. by subjecting
each of its blocksB.
to the* operation
of Theorem 1. Each relevant copyof a block G.,
which forms a blockB,
with k >z;,
furnishes a block(see (6))
of the formB*i =
1 0’
Bi3 ,
!!B.o
!! =z.,
whereBi3
is defined as in theproof
of Theorem 1. We shall show that 03B1* has
«much fewer»
(g"1)’s
than a.Indeed, letting
wehave, if B;
is embedded in withHence, applying (37)
to each of the relevantcopies
ofGk
andobserving
thatthey
are contained in(and
therefore,
in wehave,
for k=1,2,... ,
Since it follows from the definition of that
this implies
Letting
k tend toinfinity, observing
thatE k -~ 0,
0 r~ = lim x(Gk)
x(G1 ), k1
= kby (36),
- k-~~
k =
o(
IIII ) by (31 )
and II II > II I)by definition,
we obtain the relation"
II
x _
Therefore,
lim 201420142014201420142014 *~(g-1) -
r~x (g-1 ),
and thus a*
does not have the distribution x .
noo n
Remark. Theorem 4 may be
applied
to showthat,
in the sense of Theorem2,
for any k > 1 there exists even a k-normal number a such thata
fails to besimply
normal. It suffices to consider anyg-adic
distribution measu- re x such that x(F) =
for all F withII
FII
=k,
but with x({1 j)
>0, and
to construct apair
of
numbers a,
aby
the method of Theorem 4. Forexample,
if g =2,
k =2,
we can choosewhere
bx
is the Dirac measure concentrated at x. It iseasily
seen that the measure x is invariant under thebinary
shift operatorT2,
and that anycorresponding
number a is 2-normal buta*
is not.285
REFERENCES
[1]
D.G. CHAMPERNOWNE. «The construction of decimals normal in the scale of ten».Journ.
LondonMath. Soc. 8
(1933),254-260.
[2]
A.H. COPELAND and P.ERDÖS.
«Note on normal numbers». Bull. Amer. Math. Soc. 52(1946),
857-860.
[3]
H. DAVENPORT and P.ERDÖS.
«Note on normal decimals». Canad.Journ.
Math. 4(1952),
58-63.[4]
H.KÜHNLE. «Schnell konvergierende Ziffernfolgen bei speziellen zahlentheoretischen
Transforma- tionen». Diss. Univ.Stuttgart
1978.[5]
A.G. POSTNIKOV.«Ergodic problems
in thetheory of congruences and of diophantine approxima-
tions».
(Russian). Trudy
Mat. Inst. Stekov 82(1966). Engl.
transl. Proc. Stekov Inst. Math.82,
Amer.Math. Soc. 1967.
[6] J.
VILLE. «Etudecritique
de la notion de collectif».Monogr.
desProbabilités, 3, Gauthiers-Villars,
Paris 1939.